The largest line in the 13F that Oaktree Capital Management filed on 14 August for the second quarter of 2026 is not a stock. It is a put option on the world’s largest index fund, the SPDR S&P 500 ETF Trust, newly established, with a reported value of $1.012 billion. Second place goes to a put on the Nasdaq tracker Invesco QQQ worth $652 million, increased by 55.3 percent from the prior quarter.
Together that is $1.664 billion, or 23.08 percent of the entire reported equity book. Nearly a quarter of what Howard Marks’s firm discloses in this filing consists of bets that American stock indices will fall.
That is a striking filing from an investor whose best-known line is that you cannot predict the future but you can prepare for it. It is also a filing whose most important number is misread almost everywhere.
What the $1.012 billion actually denotes
When an option position appears in a 13F, the value shown is not the money at risk. It is the market value of the shares the option references. The form asks for the value of the underlying in the value column, and for the number of underlying shares — not the number of contracts — in the quantity column.
For Oaktree’s position that means 1,355,000 shares of the S&P 500 fund, valued at roughly $747 each, giving $1.012 billion. The Nasdaq position is 885,000 shares at roughly $736. What Oaktree actually paid is the option premium. How much that was appears nowhere, but the order of magnitude can be framed: for puts with a few months to run and strikes below the current price, premiums in the low single-digit percentage range of the notional are typical. Applied to $1.664 billion of notional, the real capital committed plausibly sits in the tens of millions, not at $1.664 billion.
The distinction is not academic. It decides whether this filing reads as panic or as an insurance premium. A firm putting a quarter of its assets into short bets has made a forecast. A firm spending a low single-digit percentage of its assets on protection has bought a policy. The value column looks identical in both cases.
Two further details are also absent, and both would be decisive. The form states no strike price and no expiry. Whether Oaktree is insured against a five percent decline or a thirty percent one, whether the options expire in September or next June, cannot be derived from the filing. You know that a hedge exists. Against what, and until when, you do not know.
The second misreading: this is not Oaktree’s portfolio
The $7.21 billion in this filing is not Oaktree’s assets. It is the reportable slice of them.
Oaktree is not an equity house. The firm built its name investing in distressed bonds and corporate credit, and that remains the centre of gravity. Bonds, loans, private holdings and cash are not reportable in a 13F. What the form shows is the equity side of a business that is predominantly about debt.
That materially changes how the index puts should be read. For a pure equity fund, hedging 23 percent of the book would be a statement about stocks. For a credit investor it is something else: credit risk and equity prices fall together in stress. A manager holding a large credit book who wants to cushion its decline in a crisis does not buy credit protection, which turns expensive and illiquid precisely when it is needed. They reach for the most liquid instrument available, which is options on the major equity indices. The hedge then shows up in the equity filing even though the risk it addresses never appears there at all.
Whether that is exactly the case, the form does not say. But it is the reading that fits this firm, and it explains why the largest disclosed index hedge of the quarter belongs to a credit investor.
What Marks says in public
The connection to the memos Marks is known for is not hard to draw. He has returned repeatedly in recent months to the question of whether the market for artificial intelligence constitutes a bubble, most recently in a piece titled “Is It a Bubble?” and in another on what artificial intelligence means for asset management itself.
The figures he argues from describe a concentration that is historically unusual: AI-related stocks account for the overwhelming majority of the S&P 500’s gains, a still larger share of its earnings growth, and nearly all of the capital expenditure. Estimates for the total cost of the infrastructure build-out run into the trillions, with annual capital spending approaching half a trillion dollars. His objection is never that the technology will fail. It is that who ultimately earns a return remains unclear, and that prices which discount every possible winner simultaneously leave no margin for error.
There is no contradiction between that stance and an index hedge. It is the practical form of the claim that you can prepare without predicting. A put option is not a forecast about timing. It is a decision to cap the consequences of an event for a limited, known cost — an event one considers possible but not schedulable.
The rest of the filing
Beneath the two option lines sits an equity book that fits the picture of a cautious credit investor. The largest single stock is the shipping company TORM at $524 million, reduced by 14.7 percent. Then come natural gas producer Expand Energy at $478 million, unchanged; auto supplier Garrett Motion at $275 million, almost halved; gold producer AngloGold Ashanti at $257 million; and pharmaceutical name Indivior at $251 million. Barrick Mining was increased by 22.1 percent, the affiliated lender Oaktree Specialty Lending by 325 percent.
The number of complete exits stands out: forty-two positions from the prior quarter are gone, among them Coinbase, Nokia, Petrobras, JetBlue, Okta and Block. Also gone from the first quarter is a put on an oil and gas exploration fund worth $182 million. Hedging is therefore not a new instrument here. It was moved from a sector to the whole market — and multiplied in the process.
The treatment of Core Scientific is worth noting too. The common stock position was cut by 46.6 percent while the call option on the same company was increased by 42.9 percent. The economic exposure remains, the capital tied up falls, and the maximum loss is capped at the premium. Same logic, opposite direction.
Oaktree is not alone in this posture. In the same filing round, David Tepper’s Appaloosa disclosed a newly built put on Apple covering 835,000 shares and $242 million, against a reported book of $7.73 billion. There too, what was actually risked is not what the line says.
What a private investor can do with this
The obvious mistake would be to replicate the position. It would be expensive for three reasons.
First, a hedge is only as sensible as the thing it hedges. Oaktree is very probably protecting a credit book that does not appear in the 13F at all. Buying these puts without carrying the corresponding risk hedges nothing; it opens a short position on the equity market. Same instrument, opposite action.
Second, the time component is decisive, and it is exactly what the filing omits. Puts lose value on every day that nothing happens. A hedge held permanently costs a meaningful share of returns over the years. Institutions roll such positions with models, trading desks and pricing that are not available to retail clients.
Third, the filing is six weeks old. The reporting date was 30 June; it was submitted on 14 August. Whether the options still existed on publication day — whether they were exercised, sold, rolled or expired worthless — the document does not say.
What remains is the information itself, and that is valuable enough. A very experienced investor whose business is pricing risk in credit markets judged it worthwhile at mid-year to insure the American equity market against losses on a large scale, and expanded that insurance materially versus the prior quarter. That is a statement about the price of protection and about the perceived probability of a setback. It is not a statement about when.
A tax note for investors in Germany and Austria
Anyone drawing practical conclusions from filings like this should understand the tax treatment, because it differs from that of shares. In Germany, gains from option transactions count as investment income and fall under the flat withholding tax of 25 percent plus solidarity surcharge and, where applicable, church tax. Offsetting losses from derivative transactions was subject to a separate restriction for years; that special rule has since been removed, though the treatment in any individual case still depends on how the transaction is structured. In Austria, derivatives fall under the 27.5 percent capital gains tax, and not every instrument is subject to automatic withholding by the custodian bank.
Both are points to settle with a tax adviser before the first trade rather than after. The difference from plain equity investing lies less in the rate than in the treatment of losses, and that is precisely where hedging transactions create practical work.
The real lesson in this line
The headline for this filing writes itself: Howard Marks bets $1.66 billion against the American stock market. Every number in that sentence comes from the original document, and the sentence is still misleading.
It is misleading because the value column for options shows the value of the underlying rather than the capital committed; because the reported book covers only the equity side of a firm whose business is mostly credit; and because strike and maturity simply are not fields in the form. Three omissions that turn an insurance policy into a price forecast if you do not know about them.
Reading 13F filings therefore requires knowing which lines mean what. For shares, the value column is the market value of what is owned. For options, it is the market value of something that may never be owned. Same column, two entirely different meanings, distinguished only by a short entry in a field most screeners ignore.
Try TradingView Free for 30 Days
Plus get a $15 discount on your first subscription through this link.
